Hi guys,

Im looking for a way to verify damaged wiring on robot feedback cabling. Its 20-24awg stranded wire. There are 6 wires of identical length per axis servo- and Im looking for a way to pinpoint broken conductors.

Since often times the conductors insulation isn't damaged, using a megohmeter/insulation testing does not find the issue very often.

How can I test a length of stranded wire and pick up broken strands that might still currently be somewhat in contact unless the wire is moved/stretched?

Is there a test instrument that will detect these minor differences in resistance?

 

A time-domain reflectometer (TDR) is an electronic instrument that uses time-domain reflectometry to characterize and locate faults in metallic cables (for example, twisted pair wire or coaxial cable). It can also be used to locate discontinuities in a connector, printed circuit board, or any other electrical path.

TDR Impedance Measurements:A Foundation for Signal Integrity is a good read on TDR.

Ron

 

Thanks, that definitely looks like a precise method! I was excited until I looked up the TDR unit and saw a $38,000 pricetag.

Is there any other more common test methods that could be used to find say- Out of 6, 22ga stranded wire 10 feet in length, one wire that had all but a couple strands broken inside the insulation?

I was hoping a good megger would locate issues, but It doesn't appear to be the case in my initial tests (with a Fluke 1587)

 

Hello,

A megger can be used as isolation measurement.
A TDR can see a break or short in a cable.

Bertus

 

There are lower cost TDRs but as a rule they are pricey. The problem with a megger is the risk of damaging any circuit components. If this is just point to point wiring harness using connectors you may be able to get some mating connectors and build a simple continuity tester using an inexpensive uC and a multiplex / demultiplex type design. A basic will tell you if the point to point is correct. Less anything else added it will not show you things like crosstalk or line to line shorts.

Ron

 

Ah I see some TDR testers at affordable pricing after a second look.

Will basically all TDR testers work for stranded wire and short lengths of 3-10 feet?


This is to test disconnected cabling in a robot harness. I haven't been able to get the test results I had hoped for with a megger. There are very, very minor resistance differences because the broken strands still touch under most circumstances. When the robot moves and flexes the cabling is when the intermittent problems arise.

Do you have more info on a Uc test as you mentioned? I haven't heard of that before.

 

If you know which conductor is
(Open?) you can short allof the other conductors together and measure the capacitance from the open conductor to all the others from one end of the cable and then the other.

The ratio of the capacitance from one end to the sum of the two capacitances will be nearly the same as the distance to the break to the total length.

The capacitance measurements would need to be taken at a low enough frequency to avoid transmission line effects that could affect consustency if the measurements.

 

I'm not sure what you are trying to discover.

You're trying to find which of the six cables are intermittently open?
Or
You're trying to locate the position of the intermittent open along the length of the cable?

 

Just to be able to determine a damaged cable from a "good" one.

 

Looks like a milliohm meter might accomplish this?

A pdf from a Megger brand milliohm meter shows this exact scenario- finding wires that are failing/have failed due to constant flexing.

 

Just to be able to determine a damaged cable from a "good" one.

Connect a battery and a bulb or buzzer through the suspect cable and wiggle the cable around and see/hear if the light/buzzer stops.
If you try this with a multimeter continuity buzzer it may not work too well as there is generally a delay in the response so it may miss a brief break in the wire.

 

Perhaps a change in strategy. Find cable configuration that doesn't break. Different strand numbers and strand thickness.

You can develop a quick way to troubleshoot and repair a re-occurring problem.

Or... you can eliminate the problem.

Have you verified that broken strands are the problem? Can you find broken strands on autopsy?

 

Do you have more info on a Uc test as you mentioned? I haven't heard of that before.

I can give you a rough idea of something which may work for you. There are a number of ways to test or check a wiring harness. Most harnesses are point to point, a bundle of wires with a connector on each end. The harness might have only two wires or a hundred wires, in your case you mention 6 wires. The connectors on each end of a harness have mating connectors they mate with when the harness is put in place. So you want to test a cable harness with its connectors. Hold that thought....

I would start by using a clean sheet of paper and a sharp pencil with a good eraser. List all of the parameters you wish to test in the harness, things like point to point continuity, cross talk or line to line shorts and the list goes on. Once you know what you want to test for then develop a method to test.

When I suggested using a uC (micro-Controller) and some multiplexing here is what I had in mind. I would use a chip like the CD74HC4067, CD74HCT4067 High-Speed CMOS Logic 16-Channel Analog Multiplexer/Demultiplexer which allows bi-directional signal. Taken from the linked data sheet:
" These analog multiplexers/demultiplexers control analog voltages that may vary across the voltage supply range. They are bidirectional switches thus allowing any analog input to be used as an output and vice-versa. The switches have low “on” resistance and low “off” leakages. In addition, these devices have an enable control which when high will disable all switches to their “off” state". Additionally I can buy these chips already on a breakout board relatively inexpensive from a distributor like SparkFun Electronics for about $5.00 USD each.

So I take my cable and have mating connectors for each end. Using two chips, one on each end of my cable, I can step through each of six channels. I can apply for example 5 volts to each line and look at the other end for my 5 volts. I can use a simple micro-Controller like an Arduino Uno as my uC. which cost about $25 USD available also from Sparkfun electronics. I apply about 5 volts to each line and measure the out voltage on the other endof each line. Here is an example of using an Arduino Uno in conjunction with the CD74HC4067 , there is not much to it. The possibilities are quite a few as to what you want to test and limited only by the code. The chip can do as many as 16 lines while you only need six. The receiving end can be fed into the same Arduino using any of several ADC channels and pass/fail criteria written into the code.

This is merely a suggestion. There are dozens of ways to test a cable depending on exactly which parameters you wish to test and to what extent. The test method can be as simple or complex as you want depending on what data you want.

Ron

 

Looks like a milliohm meter might accomplish this?

Maybe. Can you disconnect the wire during testing so you don't damage a circuit?

An in-circuit capacitor tester might work. It is basically an AC ohmmeter.

 

There a several things that could work quit well for a full break, but the impending doom problem is a whole lot more difficult.

You might be able to do it with a four-lead ("Kelvin" connection) high resolution ohmmeter, provided you have measurements of the wire when intact for reference, but I doubt it. If you have say 18 of 19 strands all broken but in contact with each other right up to the break point, with the broken ends are separated by a fraction of a millimetre, or worse, with broken ends touching, the resistance difference would probably be less than you'd get with an intact wire measured at different temperatures (copper has a pretty large tempco of resistance). You can calculate what to expect in that sort of scenario to get a better idea if it might work. If you had half of the strands still intact, I think it would be pretty much impossible, but I'd certainly want to do some calculations, at the very least, before dismissing it totally. Any chance you could borrow or rent a good DMM with at least 6 digits of resolution and Kelvin connection capability & do some experiments?

A TDR is unlikely to be of any use until the break is complete. There might be a small reflection due to a transition in "skin" cross sectional area [edit] a partial break. The notion of well-defined characteristic impedance in a single conductor with an unknown bunch of adjacent conductors is a bit fanciful. If you had a multi-conductor cable in which the conductors were quite uniformly in close proximity, each to its near neighbors, and consistent in lay throughout the cable and not routed too close to metal objects, a TDR could probably detect a full break quite easily, possibly even with reasonable positional accuracy once set up for the specific type of cable - a TDR must know the propagation velocity in the conductor under test to convert time to distance.
A low-cost TDR on a relatively short cable might simply find the far end of the cable, since it is likely to produce a much greater reflection than would result from some broken strands.
A network analyzer (i.e. an electrical network analyzer, not a communication network analyzer) might be useful, but if you think TDRs are expensive you ain't seen nothin'.

 

A TDR will certainly see the end of the cable and as the length of the cable is known that provides a good calibration point. I don't know what TDRs are like now but the ones I worked with had a screen which showed the response along the cable - a large reflection from the end of the cable and a smaller reflection from any discontinuities along the way. For instance you could see where a coax cable had been deformed without it being open or shorted.

 

A TDR will have a hard time finding the end of a conductor broken in the middle, but with the situation described other conductors would probably serve quite adequately for cal'ing.
The characteristics of decent coax are going to be remarkably neat and clean relative to a couple of conductors run in a common jacket with other conductors.

I haven't used a recently-made TDR. I don't even know who is making them anymore.
The screen required human interpretation, but it didn't hide things. Since the CRT is dead, I wonder about how well small reflections would show up through the digitizing noise floor.

For something like this, you might be able to construct a usable TDR using some of the ultrafast logic devices that are around to generate a (longish period) rectangular waveform with ultrafast rise or fall time and look at things with an oscilloscope. Back in the good old days a tunnel diode would be used in the pulser, but I don't think they are being made anywhere anymore. Last time I looked, several years back, I think someone in Russia or Ukraine was selling some on ebay.

[edit] For locating a full break, I think the method Dick suggests might work quite well, again rather depending on the construction of the cable. You can also detect a full break fairly well if you have access to the full length of the cable by using the inexpensive pair of tools used by telephone techs - put a square wave on the wire and use the non-contact pickup tool.

 

The ones I worked on used an avalanche pulse generator, something like this:
http://www.aholme.co.uk/Avalanche/Avalanche.htm

 

To answer some of the questions as to why and what-

My factory has 180+ robots, many of which were pretty old/used when installed. The aging, tiny feedback cabling from the resolvers is the root of many breakdown issues that are troubleshot incorrectly. Replacement harness's are about $3,500 a piece, so finding a method that can pinpoint the issues correctly everytime is important. As it is now, maintenance techs change servo's, drives and other components mainly on a whim with no proper test equipment or direction.

The cables are easy to detach, so I'm testing a simple 22-24ga wire open ended- no other components in the way. However they are not fully visible, as most of the length is run internally and a bit of a pain to remove- as it is one complete harness.

I've setup a servo test station and work instructions on how to verify problems with the servo/resolver itself, but I have not been able to find the proper test equipment to fully check the cabling correctly as of yet.


I ordered a cheap milli-ohm meter to test out, and have high hopes. I also have a friend who has a TDR that is willing to let me borrow it for awhile to try. I really feel a milli-ohm meter should be able to detect damages in the strands of the cabling in question. I hope that works out, Otherwise I plan on investigating a couple of the methods above further.

I appreciate all the advice so far!

 

There a several things that could work quit well for a full break, but the impending doom problem is a whole lot more difficult.

'.

I was hoping that with 6 identical length's per cable, I would be able to compare with the others to help pinpoint the bad.

I believe all strands are broken when the problems occur. On occasions where the worn cable is visible externally, the problem wires are easy to find as light pulling stretches the insulation on the fully broken wires.

The main issue is the wires break from constant flexing, so they come "apart", but are held mostly together- enough to work- by the insulation until the cable is flexed by a certain move of the robot.